PreCoG speeds up video editing by previewing and reusing work
PreCoG: Preview-Edit for Efficient Video Correction in Generation-Time
Computer Vision and Pattern Recognition
Summary
Creating videos from text often needs multiple tries to fix mistakes, which can take a lot of time. This paper presents PreCoG, a method that first makes quick previews to spot problems and then edits the video without starting over. By reusing earlier work, PreCoG saves computing time and keeps already-correct parts unchanged. The authors show that PreCoG produces better or similar video quality in less time compared to previous methods.
What this means in practice
- •For video content creators: Improve efficiency by correcting video generation errors without full re-rendering, reducing editing time and preserving video quality.
- •For game developers: Generate and iteratively refine game cinematic videos faster by previewing and editing partial results instead of regenerating entire clips.
Authors
Yilai Liu, Wenqi Pei, Zhifeng Lu, Shiyuan Zhang, Zongxi Liu, Hongyang Du
Abstract
Iterative text-to-video generation relies on feedback to correct compositional and temporal failures. However, existing methods typically require another full generation run for each correction, repeating computation and potentially disrupting already correct content. We introduce PreCoG, an iterative framework for correcting semantic failures during video generation. PreQA diagnoses failures from low-cost previews and constructs corrective conditions, while PreEdit edits cached generation trajectories under these conditions, attenuating orthogonal changes to limit disruption of already correct content. PreCoG completes generation only for the selected candidate, reusing its cached trajectory to reduce redundant computation and help preserve semantic consistency between the preview and the final video. Experiments show that PreCoG achieves the highest VBench and T2V-CompBench scores and approaches VSR's leading VS2 score at $0.56\times$ its inference time. It also outperforms GENMAC on all three quality and alignment metrics at $0.86\times$ its inference time.